Human immunodeficiency virus-1 (HIV) viral load In chronic HIV infection, ongoing HIV replication leads to gradual destruction of the immune system and acquired immunodeficiency syndrome (AIDS). The amount of HIV circulating in serum or plasma (viral load) reflects HIV replication throughout the lymphoid system and is a strong predictor of disease progression. The Viral Epidemiology Branch's cohort studies of HIV-infected hemophilia patients and homosexual men provide a valuable resource for studying predictors of HIV disease progression. An area of active research is the use of statistical methods to model the complex relationship between markers, such as HIV viral load, and disease progression, to better understand pathogenesis. We have recently examined the relationship between HIV viral load and disease progression in persons with late-stage HIV infection. We studied 389 HIV-infected subjects from the Multicenter Hemophilia Cohort Study who had CD4 counts below 200 cells/mm3. For these persons with advanced infection, HIV viral load was a strong predictor of AIDS risk (hazard ratio 2.37 per log10 increase in viral load). Interestingly, this effect was most pronounced immediately after the HIV viral load was measured, and it attenuated over time, implying that viral load reflects short term risk and thus, implicitly, immune deficiency. Indeed, viral load remained predictive of short term risk even after controlling for CD4 count, an established measure of immune function. These findings have implications for the clinical management of HIV-infected patients, for whom decisions regarding prophylaxis, treatment, and diagnostic evaluation often involve assessment of immune status. HIV viral load is extremely high during the first few months of infection and subsequently declines to a "setpoint" with development of HIV-specific immune responses. We studied viral load trends over time among HIV-infected children with hemophilia. On average, at two years after HIV infection these children have lower viral load "setpoints" than vertically infected children, suggesting that the age of acquisition of infection (and the maturity of the immune system) may influence how well individuals control HIV replication. On average viral load increased slowly over time. Importantly, we found significant heterogeneity among children in the level of viral load and its rate of change over time. These differences may be due to genetic variability among children or to differences in the strains infecting them. We have developed statistical methods for characterizing this heterogeneity and appropriately modeling differences among subjects in longitudinal studies of marker data. Human herpesvirus 8 (HHV-8) Advances in the understanding of the epidemiology of HHV-8 (the causative agent of Kaposi's sarcoma) depend on development of highly accurate serological tests. A focus of our recent work has been the characterization of the performance of these tests. In a recent study, we systematically evaluated five HHV-8 serological tests. Two tests (an enzyme immunoassay to the K8.1 glycoprotein and an immunofluorescence assay) appeared to be most accurate, especially in detecting low-titer antibodies. A classification tree algorithm, incorporating results from these two tests, performed better than either test alone. We have also used techniques of latent class analysis to characterize the accuracy of these test in populations from the Mediterranean region and sub-Saharan Africa. Ongoing work includes further assay development and evaluation of assay performance in epidemiological applications. HIV and cancer We are currently using the AIDS Cancer Match Registry study, which links data in U.S. AIDS and cancer registries, to study factors associated with cancer risk in HIV-infected persons. This study includes over 300,000 subjects with AIDS. Hepatitis C virus infection is common in some groups of HIV-infected persons, such as intravenous drug users and hemophiliacs. We have examined the relationship between hepatitis C virus and hepatocellular carcinoma by comparing the risk for this cancer in various HIV risk groups. Risk for hepatocellular carcinoma was highest in groups with high hepatitis C virus prevalence, suggesting that this virus causes cancer among persons with AIDS. Notably, however, even in groups with low hepatitis C virus prevalence, hepatocellular carcinoma risk was higher than in the general population. Using the AIDS Cancer Match Registry study, we have also examined the relationship between HHV-8 infection and cancer risk by looking at cancer risk in persons with Kaposi's sarcoma. Compared with other persons with AIDS, persons with AIDS-associated Kaposi's sarcoma had higher risk for certain types of non-Hodgkin's lymphoma. This relationship appears specific, since no association was found between Kaposi's sarcoma and various other hematologic and solid malignancies.